Complete excision is essential for many early stage cancers however, microscopic disease, which cannot be seen or felt by the surgeon, surrounds the main tumor and lies in draining lymph nodes, and is often left behind. These residual tumor cells increase the risk of cancer returning or spreading in almost every cancer subtype, and may increase death from cancer. Additional treatments for patients at risk for MRD are limited to empiric delivery of radiotherapy to large areas, incurring toxicity and potentially missing MRD. Despite the advent of molecular imaging agents for surgical guidance, the imagers themselves remain the limiting reagent: relatively bulky optics required for high sensitivity fluorescence imaging restrict conventional imagers from thoroughly examining small, minimally invasive tumor cavities and lymph node basins. In this proposal we introduce an entirely new platform for optical imaging -- dispensing with conventional lenses and filters for color imaging in favor of time-resolved imaging. Leveraging the unique properties of highly-efficient alloyed upconverting nanoparticles, we introduce a CMOS-based time-resolved contact imaging array platform, monolithically integrated with infrared illumination, penetrating through the silicon imager itself and deep into tissue. The delayed and upconverted emission is detected and deblurred using a custom integrated circuit, thinned to just 25 microns, with on-chip angle-selective gratings replacing focusing lenses, realizing a molecular imaging skin. Here we solve the problem of real-time intraoperative identification of MRD by introduce a thin (<200? ??m) planar molecular imaging skin to ?coat? the surface of surgical instrumentation, in essence transforming the tool itself into a microscopic imager. This ensures complete and thorough imaging of the entire complex-shaped tumor bed surface, optimizing complete resection of all disease in a single procedure addressing the issue of co-registration and sampling error?. ?Piloting this platform in a model system for breast cancer, in Aim 1 we explore the relationship between aUCNP size and biodistribution. In Aim 2, we fabricate a monolithically integrated molecular imaging skin using an IC-only with infrared through-illumination, and in Aim 3 we demonstrate our platform in a HER2+ breast cancer mouse model using an intratumoral injection of aUCNP alone, and conjugated to Trastuzumab, and anti-HER2 antibody. We choose breast cancer as microscopic residual disease is particularly prevalent and consequential: Over 25% of the 150,000 women diagnosed in the US annually with breast cancer treated with lumpectomy are found to have MRD post-operatively. MRD doubles the rate of cancer returning, from 15% to 30% over 15 years, often necessitating a second, or even third, re-excision; if left untreated, MRD could result in an additional 1,500 deaths from breast cancer annually.